New Geometry-Aware Framework Boosts Few-Shot Modulation Recognition Accuracy

Researchers have developed a new framework called Dynamic-Consistency Contrastive Learning (DyCo-CL) to improve automatic modulation recognition (AMR) in self-supervised learning. This geometry-aware approach combines Virtual Adversarial Augmentation with a semantic consistency loss, acting as an implicit spectral regularizer for more stable manifold exploration. The framework also incorporates a Signal-Adaptive Swin Backbone and a Hybrid Knowledge Fusion module to enhance representation stability and anchor them with physical priors. DyCo-CL has demonstrated a 6.27% accuracy improvement in 1-shot settings on RML benchmarks compared to existing methods. AI

IMPACT This research offers a novel approach to improve few-shot learning in signal recognition tasks, potentially enhancing performance in communication systems.

RANK_REASON The cluster contains an academic paper detailing a new method for few-shot automatic modulation recognition. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Dynamic-Consistency Contrastive Learning
• RML benchmarks
• Self-Supervised Learning
• Signal-Adaptive Swin Backbone
• Virtual Adversarial Augmentation

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →